1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799
|
"""
Create SQL statements for QuerySets.
The code in here encapsulates all of the SQL construction so that QuerySets
themselves do not have to (and could be backed by things other than SQL
databases). The abstraction barrier only works one way: this module has to know
all about the internals of models in order to get the information it needs.
"""
import copy
import difflib
import functools
import sys
from collections import Counter, namedtuple
from collections.abc import Iterator, Mapping
from itertools import chain, count, product
from string import ascii_uppercase
from django.core.exceptions import FieldDoesNotExist, FieldError
from django.db import DEFAULT_DB_ALIAS, NotSupportedError, connections
from django.db.models.aggregates import Count
from django.db.models.constants import LOOKUP_SEP
from django.db.models.expressions import (
BaseExpression,
Col,
ColPairs,
Exists,
F,
OuterRef,
RawSQL,
Ref,
ResolvedOuterRef,
Value,
)
from django.db.models.fields import Field
from django.db.models.lookups import Lookup
from django.db.models.query_utils import (
Q,
check_rel_lookup_compatibility,
refs_expression,
)
from django.db.models.sql.constants import INNER, LOUTER, ORDER_DIR, SINGLE
from django.db.models.sql.datastructures import BaseTable, Empty, Join, MultiJoin
from django.db.models.sql.where import AND, OR, ExtraWhere, NothingNode, WhereNode
from django.utils.functional import cached_property
from django.utils.regex_helper import _lazy_re_compile
from django.utils.tree import Node
__all__ = ["Query", "RawQuery"]
# Quotation marks ('"`[]), whitespace characters, semicolons, or inline
# SQL comments are forbidden in column aliases.
FORBIDDEN_ALIAS_PATTERN = _lazy_re_compile(r"['`\"\]\[;\s]|--|/\*|\*/")
# Inspired from
# https://www.postgresql.org/docs/current/sql-syntax-lexical.html#SQL-SYNTAX-IDENTIFIERS
EXPLAIN_OPTIONS_PATTERN = _lazy_re_compile(r"[\w-]+")
def get_field_names_from_opts(opts):
if opts is None:
return set()
return set(
chain.from_iterable(
(f.name, f.attname) if f.concrete else (f.name,) for f in opts.get_fields()
)
)
def get_paths_from_expression(expr):
if isinstance(expr, F):
yield expr.name
elif hasattr(expr, "flatten"):
for child in expr.flatten():
if isinstance(child, F):
yield child.name
elif isinstance(child, Q):
yield from get_children_from_q(child)
def get_children_from_q(q):
for child in q.children:
if isinstance(child, Node):
yield from get_children_from_q(child)
elif isinstance(child, tuple):
lhs, rhs = child
yield lhs
if hasattr(rhs, "resolve_expression"):
yield from get_paths_from_expression(rhs)
elif hasattr(child, "resolve_expression"):
yield from get_paths_from_expression(child)
def get_child_with_renamed_prefix(prefix, replacement, child):
from django.db.models.query import QuerySet
if isinstance(child, Node):
return rename_prefix_from_q(prefix, replacement, child)
if isinstance(child, tuple):
lhs, rhs = child
if lhs.startswith(prefix + LOOKUP_SEP):
lhs = lhs.replace(prefix, replacement, 1)
if not isinstance(rhs, F) and hasattr(rhs, "resolve_expression"):
rhs = get_child_with_renamed_prefix(prefix, replacement, rhs)
return lhs, rhs
if isinstance(child, F):
child = child.copy()
if child.name.startswith(prefix + LOOKUP_SEP):
child.name = child.name.replace(prefix, replacement, 1)
elif isinstance(child, QuerySet):
# QuerySet may contain OuterRef() references which cannot work properly
# without repointing to the filtered annotation and will spawn a
# different JOIN. Always raise ValueError instead of providing partial
# support in other cases.
raise ValueError(
"Passing a QuerySet within a FilteredRelation is not supported."
)
elif hasattr(child, "resolve_expression"):
child = child.copy()
child.set_source_expressions(
[
get_child_with_renamed_prefix(prefix, replacement, grand_child)
for grand_child in child.get_source_expressions()
]
)
return child
def rename_prefix_from_q(prefix, replacement, q):
return Q.create(
[get_child_with_renamed_prefix(prefix, replacement, c) for c in q.children],
q.connector,
q.negated,
)
JoinInfo = namedtuple(
"JoinInfo",
("final_field", "targets", "opts", "joins", "path", "transform_function"),
)
class RawQuery:
"""A single raw SQL query."""
def __init__(self, sql, using, params=()):
self.params = params
self.sql = sql
self.using = using
self.cursor = None
# Mirror some properties of a normal query so that
# the compiler can be used to process results.
self.low_mark, self.high_mark = 0, None # Used for offset/limit
self.extra_select = {}
self.annotation_select = {}
def chain(self, using):
return self.clone(using)
def clone(self, using):
return RawQuery(self.sql, using, params=self.params)
def get_columns(self):
if self.cursor is None:
self._execute_query()
converter = connections[self.using].introspection.identifier_converter
return [converter(column_meta[0]) for column_meta in self.cursor.description]
def __iter__(self):
# Always execute a new query for a new iterator.
# This could be optimized with a cache at the expense of RAM.
self._execute_query()
if not connections[self.using].features.can_use_chunked_reads:
# If the database can't use chunked reads we need to make sure we
# evaluate the entire query up front.
result = list(self.cursor)
else:
result = self.cursor
return iter(result)
def __repr__(self):
return "<%s: %s>" % (self.__class__.__name__, self)
@property
def params_type(self):
if self.params is None:
return None
return dict if isinstance(self.params, Mapping) else tuple
def __str__(self):
if self.params_type is None:
return self.sql
return self.sql % self.params_type(self.params)
def _execute_query(self):
connection = connections[self.using]
# Adapt parameters to the database, as much as possible considering
# that the target type isn't known. See #17755.
params_type = self.params_type
adapter = connection.ops.adapt_unknown_value
if params_type is tuple:
params = tuple(adapter(val) for val in self.params)
elif params_type is dict:
params = {key: adapter(val) for key, val in self.params.items()}
elif params_type is None:
params = None
else:
raise RuntimeError("Unexpected params type: %s" % params_type)
self.cursor = connection.cursor()
self.cursor.execute(self.sql, params)
ExplainInfo = namedtuple("ExplainInfo", ("format", "options"))
class Query(BaseExpression):
"""A single SQL query."""
alias_prefix = "T"
empty_result_set_value = None
subq_aliases = frozenset([alias_prefix])
compiler = "SQLCompiler"
base_table_class = BaseTable
join_class = Join
default_cols = True
default_ordering = True
standard_ordering = True
filter_is_sticky = False
subquery = False
# SQL-related attributes.
# Select and related select clauses are expressions to use in the SELECT
# clause of the query. The select is used for cases where we want to set up
# the select clause to contain other than default fields (values(),
# subqueries...). Note that annotations go to annotations dictionary.
select = ()
# The group_by attribute can have one of the following forms:
# - None: no group by at all in the query
# - A tuple of expressions: group by (at least) those expressions.
# String refs are also allowed for now.
# - True: group by all select fields of the model
# See compiler.get_group_by() for details.
group_by = None
order_by = ()
low_mark = 0 # Used for offset/limit.
high_mark = None # Used for offset/limit.
distinct = False
distinct_fields = ()
select_for_update = False
select_for_update_nowait = False
select_for_update_skip_locked = False
select_for_update_of = ()
select_for_no_key_update = False
select_related = False
# Arbitrary limit for select_related to prevents infinite recursion.
max_depth = 5
# Holds the selects defined by a call to values() or values_list()
# excluding annotation_select and extra_select.
values_select = ()
selected = None
# SQL annotation-related attributes.
annotation_select_mask = None
_annotation_select_cache = None
# Set combination attributes.
combinator = None
combinator_all = False
combined_queries = ()
# These are for extensions. The contents are more or less appended verbatim
# to the appropriate clause.
extra_select_mask = None
_extra_select_cache = None
extra_tables = ()
extra_order_by = ()
# A tuple that is a set of model field names and either True, if these are
# the fields to defer, or False if these are the only fields to load.
deferred_loading = (frozenset(), True)
explain_info = None
def __init__(self, model, alias_cols=True):
self.model = model
self.alias_refcount = {}
# alias_map is the most important data structure regarding joins.
# It's used for recording which joins exist in the query and what
# types they are. The key is the alias of the joined table (possibly
# the table name) and the value is a Join-like object (see
# sql.datastructures.Join for more information).
self.alias_map = {}
# Whether to provide alias to columns during reference resolving.
self.alias_cols = alias_cols
# Sometimes the query contains references to aliases in outer queries (as
# a result of split_exclude). Correct alias quoting needs to know these
# aliases too.
# Map external tables to whether they are aliased.
self.external_aliases = {}
self.table_map = {} # Maps table names to list of aliases.
self.used_aliases = set()
self.where = WhereNode()
# Maps alias -> Annotation Expression.
self.annotations = {}
# These are for extensions. The contents are more or less appended
# verbatim to the appropriate clause.
self.extra = {} # Maps col_alias -> (col_sql, params).
self._filtered_relations = {}
@property
def output_field(self):
if len(self.select) == 1:
select = self.select[0]
return getattr(select, "target", None) or select.field
elif len(self.annotation_select) == 1:
return next(iter(self.annotation_select.values())).output_field
@cached_property
def base_table(self):
for alias in self.alias_map:
return alias
def __str__(self):
"""
Return the query as a string of SQL with the parameter values
substituted in (use sql_with_params() to see the unsubstituted string).
Parameter values won't necessarily be quoted correctly, since that is
done by the database interface at execution time.
"""
sql, params = self.sql_with_params()
return sql % params
def sql_with_params(self):
"""
Return the query as an SQL string and the parameters that will be
substituted into the query.
"""
return self.get_compiler(DEFAULT_DB_ALIAS).as_sql()
def __deepcopy__(self, memo):
"""Limit the amount of work when a Query is deepcopied."""
result = self.clone()
memo[id(self)] = result
return result
def get_compiler(self, using=None, connection=None, elide_empty=True):
if using is None and connection is None:
raise ValueError("Need either using or connection")
if using:
connection = connections[using]
return connection.ops.compiler(self.compiler)(
self, connection, using, elide_empty
)
def get_meta(self):
"""
Return the Options instance (the model._meta) from which to start
processing. Normally, this is self.model._meta, but it can be changed
by subclasses.
"""
if self.model:
return self.model._meta
def clone(self):
"""
Return a copy of the current Query. A lightweight alternative to
deepcopy().
"""
obj = Empty()
obj.__class__ = self.__class__
# Copy references to everything.
obj.__dict__ = self.__dict__.copy()
# Clone attributes that can't use shallow copy.
obj.alias_refcount = self.alias_refcount.copy()
obj.alias_map = self.alias_map.copy()
obj.external_aliases = self.external_aliases.copy()
obj.table_map = self.table_map.copy()
obj.where = self.where.clone()
obj.annotations = self.annotations.copy()
if self.annotation_select_mask is not None:
obj.annotation_select_mask = self.annotation_select_mask.copy()
if self.combined_queries:
obj.combined_queries = tuple(
[query.clone() for query in self.combined_queries]
)
# _annotation_select_cache cannot be copied, as doing so breaks the
# (necessary) state in which both annotations and
# _annotation_select_cache point to the same underlying objects.
# It will get re-populated in the cloned queryset the next time it's
# used.
obj._annotation_select_cache = None
obj.extra = self.extra.copy()
if self.extra_select_mask is not None:
obj.extra_select_mask = self.extra_select_mask.copy()
if self._extra_select_cache is not None:
obj._extra_select_cache = self._extra_select_cache.copy()
if self.select_related is not False:
# Use deepcopy because select_related stores fields in nested
# dicts.
obj.select_related = copy.deepcopy(obj.select_related)
if "subq_aliases" in self.__dict__:
obj.subq_aliases = self.subq_aliases.copy()
obj.used_aliases = self.used_aliases.copy()
obj._filtered_relations = self._filtered_relations.copy()
# Clear the cached_property, if it exists.
obj.__dict__.pop("base_table", None)
return obj
def chain(self, klass=None):
"""
Return a copy of the current Query that's ready for another operation.
The klass argument changes the type of the Query, e.g. UpdateQuery.
"""
obj = self.clone()
if klass and obj.__class__ != klass:
obj.__class__ = klass
if not obj.filter_is_sticky:
obj.used_aliases = set()
obj.filter_is_sticky = False
if hasattr(obj, "_setup_query"):
obj._setup_query()
return obj
def relabeled_clone(self, change_map):
clone = self.clone()
clone.change_aliases(change_map)
return clone
def _get_col(self, target, field, alias):
if not self.alias_cols:
alias = None
return target.get_col(alias, field)
def get_aggregation(self, using, aggregate_exprs):
"""
Return the dictionary with the values of the existing aggregations.
"""
if not aggregate_exprs:
return {}
# Store annotation mask prior to temporarily adding aggregations for
# resolving purpose to facilitate their subsequent removal.
refs_subquery = False
refs_window = False
replacements = {}
annotation_select_mask = self.annotation_select_mask
for alias, aggregate_expr in aggregate_exprs.items():
self.check_alias(alias)
aggregate = aggregate_expr.resolve_expression(
self, allow_joins=True, reuse=None, summarize=True
)
if not aggregate.contains_aggregate:
raise TypeError("%s is not an aggregate expression" % alias)
# Temporarily add aggregate to annotations to allow remaining
# members of `aggregates` to resolve against each others.
self.append_annotation_mask([alias])
aggregate_refs = aggregate.get_refs()
refs_subquery |= any(
getattr(self.annotations[ref], "contains_subquery", False)
for ref in aggregate_refs
)
refs_window |= any(
getattr(self.annotations[ref], "contains_over_clause", True)
for ref in aggregate_refs
)
aggregate = aggregate.replace_expressions(replacements)
self.annotations[alias] = aggregate
replacements[Ref(alias, aggregate)] = aggregate
# Stash resolved aggregates now that they have been allowed to resolve
# against each other.
aggregates = {alias: self.annotations.pop(alias) for alias in aggregate_exprs}
self.set_annotation_mask(annotation_select_mask)
# Existing usage of aggregation can be determined by the presence of
# selected aggregates but also by filters against aliased aggregates.
_, having, qualify = self.where.split_having_qualify()
has_existing_aggregation = (
any(
getattr(annotation, "contains_aggregate", True)
for annotation in self.annotations.values()
)
or having
)
set_returning_annotations = {
alias
for alias, annotation in self.annotation_select.items()
if getattr(annotation, "set_returning", False)
}
# Decide if we need to use a subquery.
#
# Existing aggregations would cause incorrect results as
# get_aggregation() must produce just one result and thus must not use
# GROUP BY.
#
# If the query has limit or distinct, or uses set operations, then
# those operations must be done in a subquery so that the query
# aggregates on the limit and/or distinct results instead of applying
# the distinct and limit after the aggregation.
if (
isinstance(self.group_by, tuple)
or self.is_sliced
or has_existing_aggregation
or refs_subquery
or refs_window
or qualify
or self.distinct
or self.combinator
or set_returning_annotations
):
from django.db.models.sql.subqueries import AggregateQuery
inner_query = self.clone()
inner_query.subquery = True
outer_query = AggregateQuery(self.model, inner_query)
inner_query.select_for_update = False
inner_query.select_related = False
inner_query.set_annotation_mask(self.annotation_select)
# Queries with distinct_fields need ordering and when a limit is
# applied we must take the slice from the ordered query. Otherwise
# no need for ordering.
inner_query.clear_ordering(force=False)
if not inner_query.distinct:
# If the inner query uses default select and it has some
# aggregate annotations, then we must make sure the inner
# query is grouped by the main model's primary key. However,
# clearing the select clause can alter results if distinct is
# used.
if inner_query.default_cols and has_existing_aggregation:
inner_query.group_by = (
self.model._meta.pk.get_col(inner_query.get_initial_alias()),
)
inner_query.default_cols = False
if not qualify and not self.combinator:
# Mask existing annotations that are not referenced by
# aggregates to be pushed to the outer query unless
# filtering against window functions or if the query is
# combined as both would require complex realiasing logic.
annotation_mask = set()
if isinstance(self.group_by, tuple):
for expr in self.group_by:
annotation_mask |= expr.get_refs()
for aggregate in aggregates.values():
annotation_mask |= aggregate.get_refs()
# Avoid eliding expressions that might have an incidence on
# the implicit grouping logic.
for annotation_alias, annotation in self.annotation_select.items():
if annotation.get_group_by_cols():
annotation_mask.add(annotation_alias)
inner_query.set_annotation_mask(annotation_mask)
# Annotations that possibly return multiple rows cannot
# be masked as they might have an incidence on the query.
annotation_mask |= set_returning_annotations
# Add aggregates to the outer AggregateQuery. This requires making
# sure all columns referenced by the aggregates are selected in the
# inner query. It is achieved by retrieving all column references
# by the aggregates, explicitly selecting them in the inner query,
# and making sure the aggregates are repointed to them.
col_refs = {}
for alias, aggregate in aggregates.items():
replacements = {}
for col in self._gen_cols([aggregate], resolve_refs=False):
if not (col_ref := col_refs.get(col)):
index = len(col_refs) + 1
col_alias = f"__col{index}"
col_ref = Ref(col_alias, col)
col_refs[col] = col_ref
inner_query.add_annotation(col, col_alias)
replacements[col] = col_ref
outer_query.annotations[alias] = aggregate.replace_expressions(
replacements
)
if (
inner_query.select == ()
and not inner_query.default_cols
and not inner_query.annotation_select_mask
):
# In case of Model.objects[0:3].count(), there would be no
# field selected in the inner query, yet we must use a subquery.
# So, make sure at least one field is selected.
inner_query.select = (
self.model._meta.pk.get_col(inner_query.get_initial_alias()),
)
else:
outer_query = self
self.select = ()
self.selected = None
self.default_cols = False
self.extra = {}
if self.annotations:
# Inline reference to existing annotations and mask them as
# they are unnecessary given only the summarized aggregations
# are requested.
replacements = {
Ref(alias, annotation): annotation
for alias, annotation in self.annotations.items()
}
self.annotations = {
alias: aggregate.replace_expressions(replacements)
for alias, aggregate in aggregates.items()
}
else:
self.annotations = aggregates
self.set_annotation_mask(aggregates)
empty_set_result = [
expression.empty_result_set_value
for expression in outer_query.annotation_select.values()
]
elide_empty = not any(result is NotImplemented for result in empty_set_result)
outer_query.clear_ordering(force=True)
outer_query.clear_limits()
outer_query.select_for_update = False
outer_query.select_related = False
compiler = outer_query.get_compiler(using, elide_empty=elide_empty)
result = compiler.execute_sql(SINGLE)
if result is None:
result = empty_set_result
else:
cols = outer_query.annotation_select.values()
converters = compiler.get_converters(cols)
rows = compiler.apply_converters((result,), converters)
if compiler.has_composite_fields(cols):
rows = compiler.composite_fields_to_tuples(rows, cols)
result = next(rows)
return dict(zip(outer_query.annotation_select, result))
def get_count(self, using):
"""
Perform a COUNT() query using the current filter constraints.
"""
obj = self.clone()
return obj.get_aggregation(using, {"__count": Count("*")})["__count"]
def has_filters(self):
return self.where
def exists(self, limit=True):
q = self.clone()
if not (q.distinct and q.is_sliced):
if q.group_by is True:
q.add_fields(
(f.attname for f in self.model._meta.concrete_fields), False
)
# Disable GROUP BY aliases to avoid orphaning references to the
# SELECT clause which is about to be cleared.
q.set_group_by(allow_aliases=False)
q.clear_select_clause()
if q.combined_queries and q.combinator == "union":
q.combined_queries = tuple(
combined_query.exists(limit=False)
for combined_query in q.combined_queries
)
q.clear_ordering(force=True)
if limit is True:
q.set_limits(high=1)
q.add_annotation(Value(1), "a")
return q
def has_results(self, using):
q = self.exists()
compiler = q.get_compiler(using=using)
return compiler.has_results()
def explain(self, using, format=None, **options):
q = self.clone()
for option_name in options:
if (
not EXPLAIN_OPTIONS_PATTERN.fullmatch(option_name)
or "--" in option_name
):
raise ValueError(f"Invalid option name: {option_name!r}.")
q.explain_info = ExplainInfo(format, options)
compiler = q.get_compiler(using=using)
return "\n".join(compiler.explain_query())
def combine(self, rhs, connector):
"""
Merge the 'rhs' query into the current one (with any 'rhs' effects
being applied *after* (that is, "to the right of") anything in the
current query. 'rhs' is not modified during a call to this function.
The 'connector' parameter describes how to connect filters from the
'rhs' query.
"""
if self.model != rhs.model:
raise TypeError("Cannot combine queries on two different base models.")
if self.is_sliced:
raise TypeError("Cannot combine queries once a slice has been taken.")
if self.distinct != rhs.distinct:
raise TypeError("Cannot combine a unique query with a non-unique query.")
if self.distinct_fields != rhs.distinct_fields:
raise TypeError("Cannot combine queries with different distinct fields.")
# If lhs and rhs shares the same alias prefix, it is possible to have
# conflicting alias changes like T4 -> T5, T5 -> T6, which might end up
# as T4 -> T6 while combining two querysets. To prevent this, change an
# alias prefix of the rhs and update current aliases accordingly,
# except if the alias is the base table since it must be present in the
# query on both sides.
initial_alias = self.get_initial_alias()
rhs = rhs.clone()
rhs.bump_prefix(self, exclude={initial_alias})
# Work out how to relabel the rhs aliases, if necessary.
change_map = {}
conjunction = connector == AND
# Determine which existing joins can be reused. When combining the
# query with AND we must recreate all joins for m2m filters. When
# combining with OR we can reuse joins. The reason is that in AND
# case a single row can't fulfill a condition like:
# revrel__col=1 & revrel__col=2
# But, there might be two different related rows matching this
# condition. In OR case a single True is enough, so single row is
# enough, too.
#
# Note that we will be creating duplicate joins for non-m2m joins in
# the AND case. The results will be correct but this creates too many
# joins. This is something that could be fixed later on.
reuse = set() if conjunction else set(self.alias_map)
joinpromoter = JoinPromoter(connector, 2, False)
joinpromoter.add_votes(
j for j in self.alias_map if self.alias_map[j].join_type == INNER
)
rhs_votes = set()
# Now, add the joins from rhs query into the new query (skipping base
# table).
rhs_tables = list(rhs.alias_map)[1:]
for alias in rhs_tables:
join = rhs.alias_map[alias]
# If the left side of the join was already relabeled, use the
# updated alias.
join = join.relabeled_clone(change_map)
new_alias = self.join(join, reuse=reuse)
if join.join_type == INNER:
rhs_votes.add(new_alias)
# We can't reuse the same join again in the query. If we have two
# distinct joins for the same connection in rhs query, then the
# combined query must have two joins, too.
reuse.discard(new_alias)
if alias != new_alias:
change_map[alias] = new_alias
if not rhs.alias_refcount[alias]:
# The alias was unused in the rhs query. Unref it so that it
# will be unused in the new query, too. We have to add and
# unref the alias so that join promotion has information of
# the join type for the unused alias.
self.unref_alias(new_alias)
joinpromoter.add_votes(rhs_votes)
joinpromoter.update_join_types(self)
# Combine subqueries aliases to ensure aliases relabelling properly
# handle subqueries when combining where and select clauses.
self.subq_aliases |= rhs.subq_aliases
# Now relabel a copy of the rhs where-clause and add it to the current
# one.
w = rhs.where.clone()
w.relabel_aliases(change_map)
self.where.add(w, connector)
# Selection columns and extra extensions are those provided by 'rhs'.
if rhs.select:
self.set_select([col.relabeled_clone(change_map) for col in rhs.select])
else:
self.select = ()
if connector == OR:
# It would be nice to be able to handle this, but the queries don't
# really make sense (or return consistent value sets). Not worth
# the extra complexity when you can write a real query instead.
if self.extra and rhs.extra:
raise ValueError(
"When merging querysets using 'or', you cannot have "
"extra(select=...) on both sides."
)
self.extra.update(rhs.extra)
extra_select_mask = set()
if self.extra_select_mask is not None:
extra_select_mask.update(self.extra_select_mask)
if rhs.extra_select_mask is not None:
extra_select_mask.update(rhs.extra_select_mask)
if extra_select_mask:
self.set_extra_mask(extra_select_mask)
self.extra_tables += rhs.extra_tables
# Ordering uses the 'rhs' ordering, unless it has none, in which case
# the current ordering is used.
self.order_by = rhs.order_by or self.order_by
self.extra_order_by = rhs.extra_order_by or self.extra_order_by
def _get_defer_select_mask(self, opts, mask, select_mask=None):
if select_mask is None:
select_mask = {}
select_mask[opts.pk] = {}
# All concrete fields and related objects that are not part of the
# defer mask must be included. If a relational field is encountered it
# gets added to the mask for it be considered if `select_related` and
# the cycle continues by recursively calling this function.
for field in opts.concrete_fields + opts.related_objects:
field_mask = mask.pop(field.name, None)
field_att_mask = None
if field_attname := getattr(field, "attname", None):
field_att_mask = mask.pop(field_attname, None)
if field_mask is None and field_att_mask is None:
select_mask.setdefault(field, {})
elif field_mask:
if not field.is_relation:
raise FieldError(next(iter(field_mask)))
# Virtual fields such as many-to-many and generic foreign keys
# cannot be effectively deferred. Historically, they were
# allowed to be passed to QuerySet.defer(). Ignore such field
# references until a layer of validation at mask alteration
# time is eventually implemented.
if field.many_to_many:
continue
field_select_mask = select_mask.setdefault(field, {})
related_model = field.related_model._meta.concrete_model
self._get_defer_select_mask(
related_model._meta, field_mask, field_select_mask
)
# Remaining defer entries must be references to filtered relations
# otherwise they are surfaced as missing field errors.
for field_name, field_mask in mask.items():
if filtered_relation := self._filtered_relations.get(field_name):
relation = opts.get_field(filtered_relation.relation_name)
field_select_mask = select_mask.setdefault((field_name, relation), {})
related_model = relation.related_model._meta.concrete_model
self._get_defer_select_mask(
related_model._meta, field_mask, field_select_mask
)
else:
opts.get_field(field_name)
return select_mask
def _get_only_select_mask(self, opts, mask, select_mask=None):
if select_mask is None:
select_mask = {}
select_mask[opts.pk] = {}
# Only include fields mentioned in the mask.
for field_name, field_mask in mask.items():
field = opts.get_field(field_name)
field_select_mask = select_mask.setdefault(field, {})
if field_mask:
if not field.is_relation:
raise FieldError(next(iter(field_mask)))
related_model = field.remote_field.model._meta.concrete_model
self._get_only_select_mask(
related_model._meta, field_mask, field_select_mask
)
return select_mask
def get_select_mask(self):
"""
Convert the self.deferred_loading data structure to an alternate data
structure, describing the field that *will* be loaded. This is used to
compute the columns to select from the database and also by the
QuerySet class to work out which fields are being initialized on each
model. Models that have all their fields included aren't mentioned in
the result, only those that have field restrictions in place.
"""
field_names, defer = self.deferred_loading
if not field_names:
return {}
mask = {}
for field_name in field_names:
part_mask = mask
for part in field_name.split(LOOKUP_SEP):
part_mask = part_mask.setdefault(part, {})
opts = self.get_meta()
if defer:
return self._get_defer_select_mask(opts, mask)
return self._get_only_select_mask(opts, mask)
def table_alias(self, table_name, create=False, filtered_relation=None):
"""
Return a table alias for the given table_name and whether this is a
new alias or not.
If 'create' is true, a new alias is always created. Otherwise, the
most recently created alias for the table (if one exists) is reused.
"""
alias_list = self.table_map.get(table_name)
if not create and alias_list:
alias = alias_list[0]
self.alias_refcount[alias] += 1
return alias, False
# Create a new alias for this table.
if alias_list:
alias = "%s%d" % (self.alias_prefix, len(self.alias_map) + 1)
alias_list.append(alias)
else:
# The first occurrence of a table uses the table name directly.
alias = (
filtered_relation.alias if filtered_relation is not None else table_name
)
self.table_map[table_name] = [alias]
self.alias_refcount[alias] = 1
return alias, True
def ref_alias(self, alias):
"""Increases the reference count for this alias."""
self.alias_refcount[alias] += 1
def unref_alias(self, alias, amount=1):
"""Decreases the reference count for this alias."""
self.alias_refcount[alias] -= amount
def promote_joins(self, aliases):
"""
Promote recursively the join type of given aliases and its children to
an outer join. If 'unconditional' is False, only promote the join if
it is nullable or the parent join is an outer join.
The children promotion is done to avoid join chains that contain a LOUTER
b INNER c. So, if we have currently a INNER b INNER c and a->b is promoted,
then we must also promote b->c automatically, or otherwise the promotion
of a->b doesn't actually change anything in the query results.
"""
aliases = list(aliases)
while aliases:
alias = aliases.pop(0)
if self.alias_map[alias].join_type is None:
# This is the base table (first FROM entry) - this table
# isn't really joined at all in the query, so we should not
# alter its join type.
continue
# Only the first alias (skipped above) should have None join_type
assert self.alias_map[alias].join_type is not None
parent_alias = self.alias_map[alias].parent_alias
parent_louter = (
parent_alias and self.alias_map[parent_alias].join_type == LOUTER
)
already_louter = self.alias_map[alias].join_type == LOUTER
if (self.alias_map[alias].nullable or parent_louter) and not already_louter:
self.alias_map[alias] = self.alias_map[alias].promote()
# Join type of 'alias' changed, so re-examine all aliases that
# refer to this one.
aliases.extend(
join
for join in self.alias_map
if self.alias_map[join].parent_alias == alias
and join not in aliases
)
def demote_joins(self, aliases):
"""
Change join type from LOUTER to INNER for all joins in aliases.
Similarly to promote_joins(), this method must ensure no join chains
containing first an outer, then an inner join are generated. If we
are demoting b->c join in chain a LOUTER b LOUTER c then we must
demote a->b automatically, or otherwise the demotion of b->c doesn't
actually change anything in the query results. .
"""
aliases = list(aliases)
while aliases:
alias = aliases.pop(0)
if self.alias_map[alias].join_type == LOUTER:
self.alias_map[alias] = self.alias_map[alias].demote()
parent_alias = self.alias_map[alias].parent_alias
if self.alias_map[parent_alias].join_type == INNER:
aliases.append(parent_alias)
def reset_refcounts(self, to_counts):
"""
Reset reference counts for aliases so that they match the value passed
in `to_counts`.
"""
for alias, cur_refcount in self.alias_refcount.copy().items():
unref_amount = cur_refcount - to_counts.get(alias, 0)
self.unref_alias(alias, unref_amount)
def change_aliases(self, change_map):
"""
Change the aliases in change_map (which maps old-alias -> new-alias),
relabelling any references to them in select columns and the where
clause.
"""
if not change_map:
return self
# If keys and values of change_map were to intersect, an alias might be
# updated twice (e.g. T4 -> T5, T5 -> T6, so also T4 -> T6) depending
# on their order in change_map.
assert set(change_map).isdisjoint(change_map.values())
# 1. Update references in "select" (normal columns plus aliases),
# "group by" and "where".
self.where.relabel_aliases(change_map)
if isinstance(self.group_by, tuple):
self.group_by = tuple(
[col.relabeled_clone(change_map) for col in self.group_by]
)
self.select = tuple([col.relabeled_clone(change_map) for col in self.select])
self.annotations = self.annotations and {
key: col.relabeled_clone(change_map)
for key, col in self.annotations.items()
}
# 2. Rename the alias in the internal table/alias datastructures.
for old_alias, new_alias in change_map.items():
if old_alias not in self.alias_map:
continue
alias_data = self.alias_map[old_alias].relabeled_clone(change_map)
self.alias_map[new_alias] = alias_data
self.alias_refcount[new_alias] = self.alias_refcount[old_alias]
del self.alias_refcount[old_alias]
del self.alias_map[old_alias]
table_aliases = self.table_map[alias_data.table_name]
for pos, alias in enumerate(table_aliases):
if alias == old_alias:
table_aliases[pos] = new_alias
break
# 3. Rename the direct external aliases and the ones of combined
# queries (union, intersection, difference).
self.external_aliases = {
# Table is aliased or it's being changed and thus is aliased.
change_map.get(alias, alias): (aliased or alias in change_map)
for alias, aliased in self.external_aliases.items()
}
for combined_query in self.combined_queries:
external_change_map = {
alias: aliased
for alias, aliased in change_map.items()
if alias in combined_query.external_aliases
}
combined_query.change_aliases(external_change_map)
def bump_prefix(self, other_query, exclude=None):
"""
Change the alias prefix to the next letter in the alphabet in a way
that the other query's aliases and this query's aliases will not
conflict. Even tables that previously had no alias will get an alias
after this call. To prevent changing aliases use the exclude parameter.
"""
def prefix_gen():
"""
Generate a sequence of characters in alphabetical order:
-> 'A', 'B', 'C', ...
When the alphabet is finished, the sequence will continue with the
Cartesian product:
-> 'AA', 'AB', 'AC', ...
"""
alphabet = ascii_uppercase
prefix = chr(ord(self.alias_prefix) + 1)
yield prefix
for n in count(1):
seq = alphabet[alphabet.index(prefix) :] if prefix else alphabet
for s in product(seq, repeat=n):
yield "".join(s)
prefix = None
if self.alias_prefix != other_query.alias_prefix:
# No clashes between self and outer query should be possible.
return
# Explicitly avoid infinite loop. The constant divider is based on how
# much depth recursive subquery references add to the stack. This value
# might need to be adjusted when adding or removing function calls from
# the code path in charge of performing these operations.
local_recursion_limit = sys.getrecursionlimit() // 16
for pos, prefix in enumerate(prefix_gen()):
if prefix not in self.subq_aliases:
self.alias_prefix = prefix
break
if pos > local_recursion_limit:
raise RecursionError(
"Maximum recursion depth exceeded: too many subqueries."
)
self.subq_aliases = self.subq_aliases.union([self.alias_prefix])
other_query.subq_aliases = other_query.subq_aliases.union(self.subq_aliases)
if exclude is None:
exclude = {}
self.change_aliases(
{
alias: "%s%d" % (self.alias_prefix, pos)
for pos, alias in enumerate(self.alias_map)
if alias not in exclude
}
)
def get_initial_alias(self):
"""
Return the first alias for this query, after increasing its reference
count.
"""
if self.alias_map:
alias = self.base_table
self.ref_alias(alias)
elif self.model:
alias = self.join(self.base_table_class(self.get_meta().db_table, None))
else:
alias = None
return alias
def count_active_tables(self):
"""
Return the number of tables in this query with a non-zero reference
count. After execution, the reference counts are zeroed, so tables
added in compiler will not be seen by this method.
"""
return len([1 for count in self.alias_refcount.values() if count])
def join(self, join, reuse=None):
"""
Return an alias for the 'join', either reusing an existing alias for
that join or creating a new one. 'join' is either a base_table_class or
join_class.
The 'reuse' parameter can be either None which means all joins are
reusable, or it can be a set containing the aliases that can be reused.
A join is always created as LOUTER if the lhs alias is LOUTER to make
sure chains like t1 LOUTER t2 INNER t3 aren't generated. All new
joins are created as LOUTER if the join is nullable.
"""
reuse_aliases = [
a
for a, j in self.alias_map.items()
if (reuse is None or a in reuse) and j == join
]
if reuse_aliases:
if join.table_alias in reuse_aliases:
reuse_alias = join.table_alias
else:
# Reuse the most recent alias of the joined table
# (a many-to-many relation may be joined multiple times).
reuse_alias = reuse_aliases[-1]
self.ref_alias(reuse_alias)
return reuse_alias
# No reuse is possible, so we need a new alias.
alias, _ = self.table_alias(
join.table_name, create=True, filtered_relation=join.filtered_relation
)
if join.join_type:
if self.alias_map[join.parent_alias].join_type == LOUTER or join.nullable:
join_type = LOUTER
else:
join_type = INNER
join.join_type = join_type
join.table_alias = alias
self.alias_map[alias] = join
if filtered_relation := join.filtered_relation:
resolve_reuse = reuse
if resolve_reuse is not None:
resolve_reuse = set(reuse) | {alias}
joins_len = len(self.alias_map)
join.filtered_relation = filtered_relation.resolve_expression(
self, reuse=resolve_reuse
)
# Some joins were during expression resolving, they must be present
# before the one we just added.
if joins_len < len(self.alias_map):
self.alias_map[alias] = self.alias_map.pop(alias)
return alias
def join_parent_model(self, opts, model, alias, seen):
"""
Make sure the given 'model' is joined in the query. If 'model' isn't
a parent of 'opts' or if it is None this method is a no-op.
The 'alias' is the root alias for starting the join, 'seen' is a dict
of model -> alias of existing joins. It must also contain a mapping
of None -> some alias. This will be returned in the no-op case.
"""
if model in seen:
return seen[model]
chain = opts.get_base_chain(model)
if not chain:
return alias
curr_opts = opts
for int_model in chain:
if int_model in seen:
curr_opts = int_model._meta
alias = seen[int_model]
continue
# Proxy model have elements in base chain
# with no parents, assign the new options
# object and skip to the next base in that
# case
if not curr_opts.parents[int_model]:
curr_opts = int_model._meta
continue
link_field = curr_opts.get_ancestor_link(int_model)
join_info = self.setup_joins([link_field.name], curr_opts, alias)
curr_opts = int_model._meta
alias = seen[int_model] = join_info.joins[-1]
return alias or seen[None]
def check_alias(self, alias):
if FORBIDDEN_ALIAS_PATTERN.search(alias):
raise ValueError(
"Column aliases cannot contain whitespace characters, quotation marks, "
"semicolons, or SQL comments."
)
def add_annotation(self, annotation, alias, select=True):
"""Add a single annotation expression to the Query."""
self.check_alias(alias)
annotation = annotation.resolve_expression(self, allow_joins=True, reuse=None)
if select:
self.append_annotation_mask([alias])
else:
self.set_annotation_mask(set(self.annotation_select).difference({alias}))
self.annotations[alias] = annotation
if select and self.selected:
self.selected[alias] = alias
@property
def _subquery_fields_len(self):
if self.has_select_fields:
return sum(
len(self.model._meta.pk_fields) if field == "pk" else 1
for field in self.selected
)
return len(self.model._meta.pk_fields)
def resolve_expression(self, query, *args, **kwargs):
clone = self.clone()
# Subqueries need to use a different set of aliases than the outer query.
clone.bump_prefix(query)
clone.subquery = True
clone.where.resolve_expression(query, *args, **kwargs)
# Resolve combined queries.
if clone.combinator:
clone.combined_queries = tuple(
[
combined_query.resolve_expression(query, *args, **kwargs)
for combined_query in clone.combined_queries
]
)
for key, value in clone.annotations.items():
resolved = value.resolve_expression(query, *args, **kwargs)
if hasattr(resolved, "external_aliases"):
resolved.external_aliases.update(clone.external_aliases)
clone.annotations[key] = resolved
# Outer query's aliases are considered external.
for alias, table in query.alias_map.items():
clone.external_aliases[alias] = (
isinstance(table, Join)
and table.join_field.related_model._meta.db_table != alias
) or (
isinstance(table, BaseTable) and table.table_name != table.table_alias
)
return clone
def get_external_cols(self):
exprs = chain(self.annotations.values(), self.where.children)
return [
col
for col in self._gen_cols(exprs, include_external=True)
if col.alias in self.external_aliases
]
def get_group_by_cols(self, wrapper=None):
# If wrapper is referenced by an alias for an explicit GROUP BY through
# values() a reference to this expression and not the self must be
# returned to ensure external column references are not grouped against
# as well.
external_cols = self.get_external_cols()
if any(col.possibly_multivalued for col in external_cols):
return [wrapper or self]
return external_cols
def as_sql(self, compiler, connection):
# Some backends (e.g. Oracle) raise an error when a subquery contains
# unnecessary ORDER BY clause.
if (
self.subquery
and not connection.features.ignores_unnecessary_order_by_in_subqueries
):
self.clear_ordering(force=False)
for query in self.combined_queries:
query.clear_ordering(force=False)
sql, params = self.get_compiler(connection=connection).as_sql()
if self.subquery:
sql = "(%s)" % sql
return sql, params
def resolve_lookup_value(self, value, can_reuse, allow_joins, summarize=False):
if hasattr(value, "resolve_expression"):
value = value.resolve_expression(
self,
reuse=can_reuse,
allow_joins=allow_joins,
summarize=summarize,
)
elif isinstance(value, (list, tuple)):
# The items of the iterable may be expressions and therefore need
# to be resolved independently.
values = (
self.resolve_lookup_value(sub_value, can_reuse, allow_joins, summarize)
for sub_value in value
)
type_ = type(value)
if hasattr(type_, "_make"): # namedtuple
return type_(*values)
return type_(values)
return value
def solve_lookup_type(self, lookup, summarize=False):
"""
Solve the lookup type from the lookup (e.g.: 'foobar__id__icontains').
"""
lookup_splitted = lookup.split(LOOKUP_SEP)
if self.annotations:
annotation, expression_lookups = refs_expression(
lookup_splitted, self.annotations
)
if annotation:
expression = self.annotations[annotation]
if summarize:
expression = Ref(annotation, expression)
return expression_lookups, (), expression
_, field, _, lookup_parts = self.names_to_path(lookup_splitted, self.get_meta())
field_parts = lookup_splitted[0 : len(lookup_splitted) - len(lookup_parts)]
if len(lookup_parts) > 1 and not field_parts:
raise FieldError(
'Invalid lookup "%s" for model %s".'
% (lookup, self.get_meta().model.__name__)
)
return lookup_parts, field_parts, False
def check_query_object_type(self, value, opts, field):
"""
Check whether the object passed while querying is of the correct type.
If not, raise a ValueError specifying the wrong object.
"""
if hasattr(value, "_meta"):
if not check_rel_lookup_compatibility(value._meta.model, opts, field):
raise ValueError(
'Cannot query "%s": Must be "%s" instance.'
% (value, opts.object_name)
)
def check_related_objects(self, field, value, opts):
"""Check the type of object passed to query relations."""
if field.is_relation:
# Check that the field and the queryset use the same model in a
# query like .filter(author=Author.objects.all()). For example, the
# opts would be Author's (from the author field) and value.model
# would be Author.objects.all() queryset's .model (Author also).
# The field is the related field on the lhs side.
if (
isinstance(value, Query)
and not value.has_select_fields
and not check_rel_lookup_compatibility(value.model, opts, field)
):
raise ValueError(
'Cannot use QuerySet for "%s": Use a QuerySet for "%s".'
% (value.model._meta.object_name, opts.object_name)
)
elif hasattr(value, "_meta"):
self.check_query_object_type(value, opts, field)
elif hasattr(value, "__iter__"):
for v in value:
self.check_query_object_type(v, opts, field)
def check_filterable(self, expression):
"""Raise an error if expression cannot be used in a WHERE clause."""
if hasattr(expression, "resolve_expression") and not getattr(
expression, "filterable", True
):
raise NotSupportedError(
expression.__class__.__name__ + " is disallowed in the filter "
"clause."
)
if hasattr(expression, "get_source_expressions"):
for expr in expression.get_source_expressions():
self.check_filterable(expr)
def build_lookup(self, lookups, lhs, rhs):
"""
Try to extract transforms and lookup from given lhs.
The lhs value is something that works like SQLExpression.
The rhs value is what the lookup is going to compare against.
The lookups is a list of names to extract using get_lookup()
and get_transform().
"""
# __exact is the default lookup if one isn't given.
*transforms, lookup_name = lookups or ["exact"]
for name in transforms:
lhs = self.try_transform(lhs, name, lookups)
# First try get_lookup() so that the lookup takes precedence if the lhs
# supports both transform and lookup for the name.
lookup_class = lhs.get_lookup(lookup_name)
if not lookup_class:
# A lookup wasn't found. Try to interpret the name as a transform
# and do an Exact lookup against it.
lhs = self.try_transform(lhs, lookup_name)
lookup_name = "exact"
lookup_class = lhs.get_lookup(lookup_name)
if not lookup_class:
return
lookup = lookup_class(lhs, rhs)
# Interpret '__exact=None' as the sql 'is NULL'; otherwise, reject all
# uses of None as a query value unless the lookup supports it.
if lookup.rhs is None and not lookup.can_use_none_as_rhs:
if lookup_name not in ("exact", "iexact"):
raise ValueError("Cannot use None as a query value")
return lhs.get_lookup("isnull")(lhs, True)
# For Oracle '' is equivalent to null. The check must be done at this
# stage because join promotion can't be done in the compiler. Using
# DEFAULT_DB_ALIAS isn't nice but it's the best that can be done here.
# A similar thing is done in is_nullable(), too.
if (
lookup_name == "exact"
and lookup.rhs == ""
and connections[DEFAULT_DB_ALIAS].features.interprets_empty_strings_as_nulls
):
return lhs.get_lookup("isnull")(lhs, True)
return lookup
def try_transform(self, lhs, name, lookups=None):
"""
Helper method for build_lookup(). Try to fetch and initialize
a transform for name parameter from lhs.
"""
transform_class = lhs.get_transform(name)
if transform_class:
return transform_class(lhs)
else:
output_field = lhs.output_field.__class__
suggested_lookups = difflib.get_close_matches(
name, lhs.output_field.get_lookups()
)
if suggested_lookups:
suggestion = ", perhaps you meant %s?" % " or ".join(suggested_lookups)
else:
suggestion = "."
if lookups is not None:
name_index = lookups.index(name)
unsupported_lookup = LOOKUP_SEP.join(lookups[name_index:])
else:
unsupported_lookup = name
raise FieldError(
"Unsupported lookup '%s' for %s or join on the field not "
"permitted%s" % (unsupported_lookup, output_field.__name__, suggestion)
)
def build_filter(
self,
filter_expr,
branch_negated=False,
current_negated=False,
can_reuse=None,
allow_joins=True,
split_subq=True,
check_filterable=True,
summarize=False,
update_join_types=True,
):
"""
Build a WhereNode for a single filter clause but don't add it
to this Query. Query.add_q() will then add this filter to the where
Node.
The 'branch_negated' tells us if the current branch contains any
negations. This will be used to determine if subqueries are needed.
The 'current_negated' is used to determine if the current filter is
negated or not and this will be used to determine if IS NULL filtering
is needed.
The difference between current_negated and branch_negated is that
branch_negated is set on first negation, but current_negated is
flipped for each negation.
Note that add_filter will not do any negating itself, that is done
upper in the code by add_q().
The 'can_reuse' is a set of reusable joins for multijoins.
The method will create a filter clause that can be added to the current
query. However, if the filter isn't added to the query then the caller
is responsible for unreffing the joins used.
"""
if isinstance(filter_expr, dict):
raise FieldError("Cannot parse keyword query as dict")
if isinstance(filter_expr, Q):
return self._add_q(
filter_expr,
branch_negated=branch_negated,
current_negated=current_negated,
used_aliases=can_reuse,
allow_joins=allow_joins,
split_subq=split_subq,
check_filterable=check_filterable,
summarize=summarize,
update_join_types=update_join_types,
)
if hasattr(filter_expr, "resolve_expression"):
if not getattr(filter_expr, "conditional", False):
raise TypeError("Cannot filter against a non-conditional expression.")
condition = filter_expr.resolve_expression(
self, allow_joins=allow_joins, reuse=can_reuse, summarize=summarize
)
if not isinstance(condition, Lookup):
condition = self.build_lookup(["exact"], condition, True)
return WhereNode([condition], connector=AND), []
arg, value = filter_expr
if not arg:
raise FieldError("Cannot parse keyword query %r" % arg)
lookups, parts, reffed_expression = self.solve_lookup_type(arg, summarize)
if check_filterable:
self.check_filterable(reffed_expression)
if not allow_joins and len(parts) > 1:
raise FieldError("Joined field references are not permitted in this query")
pre_joins = self.alias_refcount.copy()
value = self.resolve_lookup_value(value, can_reuse, allow_joins, summarize)
used_joins = {
k for k, v in self.alias_refcount.items() if v > pre_joins.get(k, 0)
}
if check_filterable:
self.check_filterable(value)
if reffed_expression:
condition = self.build_lookup(lookups, reffed_expression, value)
return WhereNode([condition], connector=AND), []
opts = self.get_meta()
alias = self.get_initial_alias()
allow_many = not branch_negated or not split_subq
try:
join_info = self.setup_joins(
parts,
opts,
alias,
can_reuse=can_reuse,
allow_many=allow_many,
)
# Prevent iterator from being consumed by check_related_objects()
if isinstance(value, Iterator):
value = list(value)
self.check_related_objects(join_info.final_field, value, join_info.opts)
# split_exclude() needs to know which joins were generated for the
# lookup parts
self._lookup_joins = join_info.joins
except MultiJoin as e:
return self.split_exclude(filter_expr, can_reuse, e.names_with_path)
# Update used_joins before trimming since they are reused to determine
# which joins could be later promoted to INNER.
used_joins.update(join_info.joins)
targets, alias, join_list = self.trim_joins(
join_info.targets, join_info.joins, join_info.path
)
if can_reuse is not None:
can_reuse.update(join_list)
if join_info.final_field.is_relation:
if len(targets) == 1:
col = self._get_col(targets[0], join_info.final_field, alias)
else:
col = ColPairs(alias, targets, join_info.targets, join_info.final_field)
else:
col = self._get_col(targets[0], join_info.final_field, alias)
condition = self.build_lookup(lookups, col, value)
lookup_type = condition.lookup_name
clause = WhereNode([condition], connector=AND)
require_outer = (
lookup_type == "isnull" and condition.rhs is True and not current_negated
)
if (
current_negated
and (lookup_type != "isnull" or condition.rhs is False)
and condition.rhs is not None
):
require_outer = True
if lookup_type != "isnull":
# The condition added here will be SQL like this:
# NOT (col IS NOT NULL), where the first NOT is added in
# upper layers of code. The reason for addition is that if col
# is null, then col != someval will result in SQL "unknown"
# which isn't the same as in Python. The Python None handling
# is wanted, and it can be gotten by
# (col IS NULL OR col != someval)
# <=>
# NOT (col IS NOT NULL AND col = someval).
if (
self.is_nullable(targets[0])
or self.alias_map[join_list[-1]].join_type == LOUTER
):
lookup_class = targets[0].get_lookup("isnull")
col = self._get_col(targets[0], join_info.targets[0], alias)
clause.add(lookup_class(col, False), AND)
# If someval is a nullable column, someval IS NOT NULL is
# added.
if isinstance(value, Col) and self.is_nullable(value.target):
lookup_class = value.target.get_lookup("isnull")
clause.add(lookup_class(value, False), AND)
return clause, used_joins if not require_outer else ()
def add_filter(self, filter_lhs, filter_rhs):
self.add_q(Q((filter_lhs, filter_rhs)))
def add_q(self, q_object, reuse_all=False):
"""
A preprocessor for the internal _add_q(). Responsible for doing final
join promotion.
"""
# For join promotion this case is doing an AND for the added q_object
# and existing conditions. So, any existing inner join forces the join
# type to remain inner. Existing outer joins can however be demoted.
# (Consider case where rel_a is LOUTER and rel_a__col=1 is added - if
# rel_a doesn't produce any rows, then the whole condition must fail.
# So, demotion is OK.
existing_inner = {
a for a in self.alias_map if self.alias_map[a].join_type == INNER
}
if reuse_all:
can_reuse = set(self.alias_map)
else:
can_reuse = self.used_aliases
clause, _ = self._add_q(q_object, can_reuse)
if clause:
self.where.add(clause, AND)
self.demote_joins(existing_inner)
def build_where(self, filter_expr):
return self.build_filter(filter_expr, allow_joins=False)[0]
def clear_where(self):
self.where = WhereNode()
def _add_q(
self,
q_object,
used_aliases,
branch_negated=False,
current_negated=False,
allow_joins=True,
split_subq=True,
check_filterable=True,
summarize=False,
update_join_types=True,
):
"""Add a Q-object to the current filter."""
connector = q_object.connector
current_negated ^= q_object.negated
branch_negated = branch_negated or q_object.negated
target_clause = WhereNode(connector=connector, negated=q_object.negated)
joinpromoter = JoinPromoter(
q_object.connector, len(q_object.children), current_negated
)
for child in q_object.children:
child_clause, needed_inner = self.build_filter(
child,
can_reuse=used_aliases,
branch_negated=branch_negated,
current_negated=current_negated,
allow_joins=allow_joins,
split_subq=split_subq,
check_filterable=check_filterable,
summarize=summarize,
update_join_types=update_join_types,
)
joinpromoter.add_votes(needed_inner)
if child_clause:
target_clause.add(child_clause, connector)
if update_join_types:
needed_inner = joinpromoter.update_join_types(self)
else:
needed_inner = []
return target_clause, needed_inner
def add_filtered_relation(self, filtered_relation, alias):
filtered_relation.alias = alias
relation_lookup_parts, relation_field_parts, _ = self.solve_lookup_type(
filtered_relation.relation_name
)
if relation_lookup_parts:
raise ValueError(
"FilteredRelation's relation_name cannot contain lookups "
"(got %r)." % filtered_relation.relation_name
)
for lookup in get_children_from_q(filtered_relation.condition):
lookup_parts, lookup_field_parts, _ = self.solve_lookup_type(lookup)
shift = 2 if not lookup_parts else 1
lookup_field_path = lookup_field_parts[:-shift]
for idx, lookup_field_part in enumerate(lookup_field_path):
if len(relation_field_parts) > idx:
if relation_field_parts[idx] != lookup_field_part:
raise ValueError(
"FilteredRelation's condition doesn't support "
"relations outside the %r (got %r)."
% (filtered_relation.relation_name, lookup)
)
if len(lookup_field_parts) > len(relation_field_parts) + 1:
raise ValueError(
"FilteredRelation's condition doesn't support nested "
"relations deeper than the relation_name (got %r for "
"%r)." % (lookup, filtered_relation.relation_name)
)
filtered_relation.condition = rename_prefix_from_q(
filtered_relation.relation_name,
alias,
filtered_relation.condition,
)
self._filtered_relations[filtered_relation.alias] = filtered_relation
def names_to_path(self, names, opts, allow_many=True, fail_on_missing=False):
"""
Walk the list of names and turns them into PathInfo tuples. A single
name in 'names' can generate multiple PathInfos (m2m, for example).
'names' is the path of names to travel, 'opts' is the model Options we
start the name resolving from, 'allow_many' is as for setup_joins().
If fail_on_missing is set to True, then a name that can't be resolved
will generate a FieldError.
Return a list of PathInfo tuples. In addition return the final field
(the last used join field) and target (which is a field guaranteed to
contain the same value as the final field). Finally, return those names
that weren't found (which are likely transforms and the final lookup).
"""
path, names_with_path = [], []
for pos, name in enumerate(names):
cur_names_with_path = (name, [])
if name == "pk" and opts is not None:
name = opts.pk.name
field = None
filtered_relation = None
try:
if opts is None:
raise FieldDoesNotExist
field = opts.get_field(name)
except FieldDoesNotExist:
if name in self.annotation_select:
field = self.annotation_select[name].output_field
elif name in self._filtered_relations and pos == 0:
filtered_relation = self._filtered_relations[name]
if LOOKUP_SEP in filtered_relation.relation_name:
parts = filtered_relation.relation_name.split(LOOKUP_SEP)
filtered_relation_path, field, _, _ = self.names_to_path(
parts,
opts,
allow_many,
fail_on_missing,
)
path.extend(filtered_relation_path[:-1])
else:
field = opts.get_field(filtered_relation.relation_name)
if field is not None:
# Fields that contain one-to-many relations with a generic
# model (like a GenericForeignKey) cannot generate reverse
# relations and therefore cannot be used for reverse querying.
if field.is_relation and not field.related_model:
raise FieldError(
"Field %r does not generate an automatic reverse "
"relation and therefore cannot be used for reverse "
"querying. If it is a GenericForeignKey, consider "
"adding a GenericRelation." % name
)
try:
model = field.model._meta.concrete_model
except AttributeError:
# QuerySet.annotate() may introduce fields that aren't
# attached to a model.
model = None
else:
# We didn't find the current field, so move position back
# one step.
pos -= 1
if pos == -1 or fail_on_missing:
available = sorted(
[
*get_field_names_from_opts(opts),
*self.annotation_select,
*self._filtered_relations,
]
)
raise FieldError(
"Cannot resolve keyword '%s' into field. "
"Choices are: %s" % (name, ", ".join(available))
)
break
# Check if we need any joins for concrete inheritance cases (the
# field lives in parent, but we are currently in one of its
# children)
if opts is not None and model is not opts.model:
path_to_parent = opts.get_path_to_parent(model)
if path_to_parent:
path.extend(path_to_parent)
cur_names_with_path[1].extend(path_to_parent)
opts = path_to_parent[-1].to_opts
if hasattr(field, "path_infos"):
if filtered_relation:
pathinfos = field.get_path_info(filtered_relation)
else:
pathinfos = field.path_infos
if not allow_many:
for inner_pos, p in enumerate(pathinfos):
if p.m2m:
cur_names_with_path[1].extend(pathinfos[0 : inner_pos + 1])
names_with_path.append(cur_names_with_path)
raise MultiJoin(pos + 1, names_with_path)
last = pathinfos[-1]
path.extend(pathinfos)
final_field = last.join_field
opts = last.to_opts
targets = last.target_fields
cur_names_with_path[1].extend(pathinfos)
names_with_path.append(cur_names_with_path)
else:
# Local non-relational field.
final_field = field
targets = (field,)
if fail_on_missing and pos + 1 != len(names):
raise FieldError(
"Cannot resolve keyword %r into field. Join on '%s'"
" not permitted." % (names[pos + 1], name)
)
break
return path, final_field, targets, names[pos + 1 :]
def setup_joins(
self,
names,
opts,
alias,
can_reuse=None,
allow_many=True,
):
"""
Compute the necessary table joins for the passage through the fields
given in 'names'. 'opts' is the Options class for the current model
(which gives the table we are starting from), 'alias' is the alias for
the table to start the joining from.
The 'can_reuse' defines the reverse foreign key joins we can reuse. It
can be None in which case all joins are reusable or a set of aliases
that can be reused. Note that non-reverse foreign keys are always
reusable when using setup_joins().
If 'allow_many' is False, then any reverse foreign key seen will
generate a MultiJoin exception.
Return the final field involved in the joins, the target field (used
for any 'where' constraint), the final 'opts' value, the joins, the
field path traveled to generate the joins, and a transform function
that takes a field and alias and is equivalent to `field.get_col(alias)`
in the simple case but wraps field transforms if they were included in
names.
The target field is the field containing the concrete value. Final
field can be something different, for example foreign key pointing to
that value. Final field is needed for example in some value
conversions (convert 'obj' in fk__id=obj to pk val using the foreign
key field for example).
"""
joins = [alias]
# The transform can't be applied yet, as joins must be trimmed later.
# To avoid making every caller of this method look up transforms
# directly, compute transforms here and create a partial that converts
# fields to the appropriate wrapped version.
def final_transformer(field, alias):
if not self.alias_cols:
alias = None
return field.get_col(alias)
# Try resolving all the names as fields first. If there's an error,
# treat trailing names as lookups until a field can be resolved.
last_field_exception = None
for pivot in range(len(names), 0, -1):
try:
path, final_field, targets, rest = self.names_to_path(
names[:pivot],
opts,
allow_many,
fail_on_missing=True,
)
except FieldError as exc:
if pivot == 1:
# The first item cannot be a lookup, so it's safe
# to raise the field error here.
raise
else:
last_field_exception = exc
else:
# The transforms are the remaining items that couldn't be
# resolved into fields.
transforms = names[pivot:]
break
for name in transforms:
def transform(field, alias, *, name, previous):
try:
wrapped = previous(field, alias)
return self.try_transform(wrapped, name)
except FieldError:
# FieldError is raised if the transform doesn't exist.
if isinstance(final_field, Field) and last_field_exception:
raise last_field_exception
else:
raise
final_transformer = functools.partial(
transform, name=name, previous=final_transformer
)
final_transformer.has_transforms = True
# Then, add the path to the query's joins. Note that we can't trim
# joins at this stage - we will need the information about join type
# of the trimmed joins.
for join in path:
if join.filtered_relation:
filtered_relation = join.filtered_relation.clone()
table_alias = filtered_relation.alias
else:
filtered_relation = None
table_alias = None
opts = join.to_opts
if join.direct:
nullable = self.is_nullable(join.join_field)
else:
nullable = True
connection = self.join_class(
opts.db_table,
alias,
table_alias,
INNER,
join.join_field,
nullable,
filtered_relation=filtered_relation,
)
reuse = can_reuse if join.m2m else None
alias = self.join(connection, reuse=reuse)
joins.append(alias)
if join.filtered_relation and can_reuse is not None:
can_reuse.add(alias)
return JoinInfo(final_field, targets, opts, joins, path, final_transformer)
def trim_joins(self, targets, joins, path):
"""
The 'target' parameter is the final field being joined to, 'joins'
is the full list of join aliases. The 'path' contain the PathInfos
used to create the joins.
Return the final target field and table alias and the new active
joins.
Always trim any direct join if the target column is already in the
previous table. Can't trim reverse joins as it's unknown if there's
anything on the other side of the join.
"""
joins = joins[:]
for pos, info in enumerate(reversed(path)):
if len(joins) == 1 or not info.direct:
break
if info.filtered_relation:
break
join_targets = {t.column for t in info.join_field.foreign_related_fields}
cur_targets = {t.column for t in targets}
if not cur_targets.issubset(join_targets):
break
targets_dict = {
r[1].column: r[0]
for r in info.join_field.related_fields
if r[1].column in cur_targets
}
targets = tuple(targets_dict[t.column] for t in targets)
self.unref_alias(joins.pop())
return targets, joins[-1], joins
@classmethod
def _gen_cols(cls, exprs, include_external=False, resolve_refs=True):
for expr in exprs:
if isinstance(expr, Col):
yield expr
elif include_external and callable(
getattr(expr, "get_external_cols", None)
):
yield from expr.get_external_cols()
elif hasattr(expr, "get_source_expressions"):
if not resolve_refs and isinstance(expr, Ref):
continue
yield from cls._gen_cols(
expr.get_source_expressions(),
include_external=include_external,
resolve_refs=resolve_refs,
)
@classmethod
def _gen_col_aliases(cls, exprs):
yield from (expr.alias for expr in cls._gen_cols(exprs))
def resolve_ref(self, name, allow_joins=True, reuse=None, summarize=False):
annotation = self.annotations.get(name)
if annotation is not None:
if not allow_joins:
for alias in self._gen_col_aliases([annotation]):
if isinstance(self.alias_map[alias], Join):
raise FieldError(
"Joined field references are not permitted in this query"
)
if summarize:
# Summarize currently means we are doing an aggregate() query
# which is executed as a wrapped subquery if any of the
# aggregate() elements reference an existing annotation. In
# that case we need to return a Ref to the subquery's annotation.
if name not in self.annotation_select:
raise FieldError(
"Cannot aggregate over the '%s' alias. Use annotate() "
"to promote it." % name
)
return Ref(name, self.annotation_select[name])
else:
return annotation
else:
field_list = name.split(LOOKUP_SEP)
annotation = self.annotations.get(field_list[0])
if annotation is not None:
for transform in field_list[1:]:
annotation = self.try_transform(annotation, transform)
return annotation
join_info = self.setup_joins(
field_list, self.get_meta(), self.get_initial_alias(), can_reuse=reuse
)
targets, final_alias, join_list = self.trim_joins(
join_info.targets, join_info.joins, join_info.path
)
if not allow_joins and len(join_list) > 1:
raise FieldError(
"Joined field references are not permitted in this query"
)
if len(targets) > 1:
raise FieldError(
"Referencing multicolumn fields with F() objects isn't supported"
)
# Verify that the last lookup in name is a field or a transform:
# transform_function() raises FieldError if not.
transform = join_info.transform_function(targets[0], final_alias)
if reuse is not None:
reuse.update(join_list)
return transform
def split_exclude(self, filter_expr, can_reuse, names_with_path):
"""
When doing an exclude against any kind of N-to-many relation, we need
to use a subquery. This method constructs the nested query, given the
original exclude filter (filter_expr) and the portion up to the first
N-to-many relation field.
For example, if the origin filter is ~Q(child__name='foo'), filter_expr
is ('child__name', 'foo') and can_reuse is a set of joins usable for
filters in the original query.
We will turn this into equivalent of:
WHERE NOT EXISTS(
SELECT 1
FROM child
WHERE name = 'foo' AND child.parent_id = parent.id
LIMIT 1
)
"""
# Generate the inner query.
query = self.__class__(self.model)
query._filtered_relations = self._filtered_relations
filter_lhs, filter_rhs = filter_expr
if isinstance(filter_rhs, OuterRef):
filter_rhs = OuterRef(filter_rhs)
elif isinstance(filter_rhs, F):
filter_rhs = OuterRef(filter_rhs.name)
query.add_filter(filter_lhs, filter_rhs)
query.clear_ordering(force=True)
# Try to have as simple as possible subquery -> trim leading joins from
# the subquery.
trimmed_prefix, contains_louter = query.trim_start(names_with_path)
col = query.select[0]
select_field = col.target
alias = col.alias
if alias in can_reuse:
pk = select_field.model._meta.pk
# Need to add a restriction so that outer query's filters are in effect for
# the subquery, too.
query.bump_prefix(self)
lookup_class = select_field.get_lookup("exact")
# Note that the query.select[0].alias is different from alias
# due to bump_prefix above.
lookup = lookup_class(pk.get_col(query.select[0].alias), pk.get_col(alias))
query.where.add(lookup, AND)
query.external_aliases[alias] = True
else:
lookup_class = select_field.get_lookup("exact")
lookup = lookup_class(col, ResolvedOuterRef(trimmed_prefix))
query.where.add(lookup, AND)
condition, needed_inner = self.build_filter(Exists(query))
if contains_louter:
or_null_condition, _ = self.build_filter(
("%s__isnull" % trimmed_prefix, True),
current_negated=True,
branch_negated=True,
can_reuse=can_reuse,
)
condition.add(or_null_condition, OR)
# Note that the end result will be:
# NOT EXISTS (inner_q) OR outercol IS NULL
# this might look crazy but due to how NULL works, this seems to be
# correct. If the IS NULL check is removed, then if outercol
# IS NULL we will not match the row.
return condition, needed_inner
def set_empty(self):
self.where.add(NothingNode(), AND)
for query in self.combined_queries:
query.set_empty()
def is_empty(self):
return any(isinstance(c, NothingNode) for c in self.where.children)
def set_limits(self, low=None, high=None):
"""
Adjust the limits on the rows retrieved. Use low/high to set these,
as it makes it more Pythonic to read and write. When the SQL query is
created, convert them to the appropriate offset and limit values.
Apply any limits passed in here to the existing constraints. Add low
to the current low value and clamp both to any existing high value.
"""
if high is not None:
if self.high_mark is not None:
self.high_mark = min(self.high_mark, self.low_mark + high)
else:
self.high_mark = self.low_mark + high
if low is not None:
if self.high_mark is not None:
self.low_mark = min(self.high_mark, self.low_mark + low)
else:
self.low_mark = self.low_mark + low
if self.low_mark == self.high_mark:
self.set_empty()
def clear_limits(self):
"""Clear any existing limits."""
self.low_mark, self.high_mark = 0, None
@property
def is_sliced(self):
return self.low_mark != 0 or self.high_mark is not None
def has_limit_one(self):
return self.high_mark is not None and (self.high_mark - self.low_mark) == 1
def can_filter(self):
"""
Return True if adding filters to this instance is still possible.
Typically, this means no limits or offsets have been put on the results.
"""
return not self.is_sliced
def clear_select_clause(self):
"""Remove all fields from SELECT clause."""
self.select = ()
self.default_cols = False
self.select_related = False
self.set_extra_mask(())
self.set_annotation_mask(())
self.selected = None
def clear_select_fields(self):
"""
Clear the list of fields to select (but not extra_select columns).
Some queryset types completely replace any existing list of select
columns.
"""
self.select = ()
self.values_select = ()
self.selected = None
def add_select_col(self, col, name):
self.select += (col,)
self.values_select += (name,)
self.selected[name] = len(self.select) - 1
def set_select(self, cols):
self.default_cols = False
self.select = tuple(cols)
def add_distinct_fields(self, *field_names):
"""
Add and resolve the given fields to the query's "distinct on" clause.
"""
self.distinct_fields = field_names
self.distinct = True
def add_fields(self, field_names, allow_m2m=True):
"""
Add the given (model) fields to the select set. Add the field names in
the order specified.
"""
alias = self.get_initial_alias()
opts = self.get_meta()
try:
cols = []
for name in field_names:
# Join promotion note - we must not remove any rows here, so
# if there is no existing joins, use outer join.
join_info = self.setup_joins(
name.split(LOOKUP_SEP), opts, alias, allow_many=allow_m2m
)
targets, final_alias, joins = self.trim_joins(
join_info.targets,
join_info.joins,
join_info.path,
)
for target in targets:
cols.append(join_info.transform_function(target, final_alias))
if cols:
self.set_select(cols)
except MultiJoin:
raise FieldError("Invalid field name: '%s'" % name)
except FieldError:
if LOOKUP_SEP in name:
# For lookups spanning over relationships, show the error
# from the model on which the lookup failed.
raise
else:
names = sorted(
[
*get_field_names_from_opts(opts),
*self.extra,
*self.annotation_select,
*self._filtered_relations,
]
)
raise FieldError(
"Cannot resolve keyword %r into field. "
"Choices are: %s" % (name, ", ".join(names))
)
def add_ordering(self, *ordering):
"""
Add items from the 'ordering' sequence to the query's "order by"
clause. These items are either field names (not column names) --
possibly with a direction prefix ('-' or '?') -- or OrderBy
expressions.
If 'ordering' is empty, clear all ordering from the query.
"""
errors = []
for item in ordering:
if isinstance(item, str):
if item == "?":
continue
item = item.removeprefix("-")
if item in self.annotations:
continue
if self.extra and item in self.extra:
continue
# names_to_path() validates the lookup. A descriptive
# FieldError will be raise if it's not.
self.names_to_path(item.split(LOOKUP_SEP), self.model._meta)
elif not hasattr(item, "resolve_expression"):
errors.append(item)
if getattr(item, "contains_aggregate", False):
raise FieldError(
"Using an aggregate in order_by() without also including "
"it in annotate() is not allowed: %s" % item
)
if errors:
raise FieldError("Invalid order_by arguments: %s" % errors)
if ordering:
self.order_by += ordering
else:
self.default_ordering = False
def clear_ordering(self, force=False, clear_default=True):
"""
Remove any ordering settings if the current query allows it without
side effects, set 'force' to True to clear the ordering regardless.
If 'clear_default' is True, there will be no ordering in the resulting
query (not even the model's default).
"""
if not force and (
self.is_sliced or self.distinct_fields or self.select_for_update
):
return
self.order_by = ()
self.extra_order_by = ()
if clear_default:
self.default_ordering = False
def set_group_by(self, allow_aliases=True):
"""
Expand the GROUP BY clause required by the query.
This will usually be the set of all non-aggregate fields in the
return data. If the database backend supports grouping by the
primary key, and the query would be equivalent, the optimization
will be made automatically.
"""
if allow_aliases and self.values_select:
# If grouping by aliases is allowed assign selected value aliases
# by moving them to annotations.
group_by_annotations = {}
values_select = {}
for alias, expr in zip(self.values_select, self.select):
if isinstance(expr, Col):
values_select[alias] = expr
else:
group_by_annotations[alias] = expr
self.annotations = {**group_by_annotations, **self.annotations}
self.append_annotation_mask(group_by_annotations)
self.select = tuple(values_select.values())
self.values_select = tuple(values_select)
if self.selected is not None:
for index, value_select in enumerate(values_select):
self.selected[value_select] = index
group_by = list(self.select)
for alias, annotation in self.annotation_select.items():
if not (group_by_cols := annotation.get_group_by_cols()):
continue
if allow_aliases and not annotation.contains_aggregate:
group_by.append(Ref(alias, annotation))
else:
group_by.extend(group_by_cols)
self.group_by = tuple(group_by)
def add_select_related(self, fields):
"""
Set up the select_related data structure so that we only select
certain related models (as opposed to all models, when
self.select_related=True).
"""
if isinstance(self.select_related, bool):
field_dict = {}
else:
field_dict = self.select_related
for field in fields:
d = field_dict
for part in field.split(LOOKUP_SEP):
d = d.setdefault(part, {})
self.select_related = field_dict
def add_extra(self, select, select_params, where, params, tables, order_by):
"""
Add data to the various extra_* attributes for user-created additions
to the query.
"""
if select:
# We need to pair any placeholder markers in the 'select'
# dictionary with their parameters in 'select_params' so that
# subsequent updates to the select dictionary also adjust the
# parameters appropriately.
select_pairs = {}
if select_params:
param_iter = iter(select_params)
else:
param_iter = iter([])
for name, entry in select.items():
self.check_alias(name)
entry = str(entry)
entry_params = []
pos = entry.find("%s")
while pos != -1:
if pos == 0 or entry[pos - 1] != "%":
entry_params.append(next(param_iter))
pos = entry.find("%s", pos + 2)
select_pairs[name] = (entry, entry_params)
self.extra.update(select_pairs)
if where or params:
self.where.add(ExtraWhere(where, params), AND)
if tables:
self.extra_tables += tuple(tables)
if order_by:
self.extra_order_by = order_by
def clear_deferred_loading(self):
"""Remove any fields from the deferred loading set."""
self.deferred_loading = (frozenset(), True)
def add_deferred_loading(self, field_names):
"""
Add the given list of model field names to the set of fields to
exclude from loading from the database when automatic column selection
is done. Add the new field names to any existing field names that
are deferred (or removed from any existing field names that are marked
as the only ones for immediate loading).
"""
# Fields on related models are stored in the literal double-underscore
# format, so that we can use a set datastructure. We do the foo__bar
# splitting and handling when computing the SQL column names (as part of
# get_columns()).
existing, defer = self.deferred_loading
if defer:
# Add to existing deferred names.
self.deferred_loading = existing.union(field_names), True
else:
# Remove names from the set of any existing "immediate load" names.
if new_existing := existing.difference(field_names):
self.deferred_loading = new_existing, False
else:
self.clear_deferred_loading()
if new_only := set(field_names).difference(existing):
self.deferred_loading = new_only, True
def add_immediate_loading(self, field_names):
"""
Add the given list of model field names to the set of fields to
retrieve when the SQL is executed ("immediate loading" fields). The
field names replace any existing immediate loading field names. If
there are field names already specified for deferred loading, remove
those names from the new field_names before storing the new names
for immediate loading. (That is, immediate loading overrides any
existing immediate values, but respects existing deferrals.)
"""
existing, defer = self.deferred_loading
field_names = set(field_names)
if "pk" in field_names:
field_names.remove("pk")
field_names.add(self.get_meta().pk.name)
if defer:
# Remove any existing deferred names from the current set before
# setting the new names.
self.deferred_loading = field_names.difference(existing), False
else:
# Replace any existing "immediate load" field names.
self.deferred_loading = frozenset(field_names), False
def set_annotation_mask(self, names):
"""Set the mask of annotations that will be returned by the SELECT."""
if names is None:
self.annotation_select_mask = None
else:
self.annotation_select_mask = set(names)
if self.selected:
# Prune the masked annotations.
self.selected = {
key: value
for key, value in self.selected.items()
if not isinstance(value, str)
or value in self.annotation_select_mask
}
# Append the unmasked annotations.
for name in names:
self.selected[name] = name
self._annotation_select_cache = None
def append_annotation_mask(self, names):
if self.annotation_select_mask is not None:
self.set_annotation_mask(self.annotation_select_mask.union(names))
def set_extra_mask(self, names):
"""
Set the mask of extra select items that will be returned by SELECT.
Don't remove them from the Query since they might be used later.
"""
if names is None:
self.extra_select_mask = None
else:
self.extra_select_mask = set(names)
self._extra_select_cache = None
@property
def has_select_fields(self):
return self.selected is not None
def set_values(self, fields):
self.select_related = False
self.clear_deferred_loading()
self.clear_select_fields()
selected = {}
if fields:
for field in fields:
self.check_alias(field)
field_names = []
extra_names = []
annotation_names = []
if not self.extra and not self.annotations:
# Shortcut - if there are no extra or annotations, then
# the values() clause must be just field names.
field_names = list(fields)
selected = dict(zip(fields, range(len(fields))))
else:
self.default_cols = False
for f in fields:
if extra := self.extra_select.get(f):
extra_names.append(f)
selected[f] = RawSQL(*extra)
elif f in self.annotation_select:
annotation_names.append(f)
selected[f] = f
elif f in self.annotations:
raise FieldError(
f"Cannot select the '{f}' alias. Use annotate() to "
"promote it."
)
else:
# Call `names_to_path` to ensure a FieldError including
# annotations about to be masked as valid choices if
# `f` is not resolvable.
if self.annotation_select:
self.names_to_path(f.split(LOOKUP_SEP), self.model._meta)
selected[f] = len(field_names)
field_names.append(f)
self.set_extra_mask(extra_names)
self.set_annotation_mask(annotation_names)
else:
field_names = [f.attname for f in self.model._meta.concrete_fields]
selected = dict.fromkeys(field_names, None)
# Selected annotations must be known before setting the GROUP BY
# clause.
if self.group_by is True:
self.add_fields(
(f.attname for f in self.model._meta.concrete_fields), False
)
# Disable GROUP BY aliases to avoid orphaning references to the
# SELECT clause which is about to be cleared.
self.set_group_by(allow_aliases=False)
self.clear_select_fields()
elif self.group_by:
# Resolve GROUP BY annotation references if they are not part of
# the selected fields anymore.
group_by = []
for expr in self.group_by:
if isinstance(expr, Ref) and expr.refs not in selected:
expr = self.annotations[expr.refs]
group_by.append(expr)
self.group_by = tuple(group_by)
self.values_select = tuple(field_names)
self.add_fields(field_names, True)
self.selected = selected if fields else None
@property
def annotation_select(self):
"""
Return the dictionary of aggregate columns that are not masked and
should be used in the SELECT clause. Cache this result for performance.
"""
if self._annotation_select_cache is not None:
return self._annotation_select_cache
elif not self.annotations:
return {}
elif self.annotation_select_mask is not None:
self._annotation_select_cache = {
k: v
for k, v in self.annotations.items()
if k in self.annotation_select_mask
}
return self._annotation_select_cache
else:
return self.annotations
@property
def extra_select(self):
if self._extra_select_cache is not None:
return self._extra_select_cache
if not self.extra:
return {}
elif self.extra_select_mask is not None:
self._extra_select_cache = {
k: v for k, v in self.extra.items() if k in self.extra_select_mask
}
return self._extra_select_cache
else:
return self.extra
def trim_start(self, names_with_path):
"""
Trim joins from the start of the join path. The candidates for trim
are the PathInfos in names_with_path structure that are m2m joins.
Also set the select column so the start matches the join.
This method is meant to be used for generating the subquery joins &
cols in split_exclude().
Return a lookup usable for doing outerq.filter(lookup=self) and a
boolean indicating if the joins in the prefix contain a LEFT OUTER join.
_"""
all_paths = []
for _, paths in names_with_path:
all_paths.extend(paths)
contains_louter = False
# Trim and operate only on tables that were generated for
# the lookup part of the query. That is, avoid trimming
# joins generated for F() expressions.
lookup_tables = [
t for t in self.alias_map if t in self._lookup_joins or t == self.base_table
]
for trimmed_paths, path in enumerate(all_paths):
if path.m2m:
break
if self.alias_map[lookup_tables[trimmed_paths + 1]].join_type == LOUTER:
contains_louter = True
alias = lookup_tables[trimmed_paths]
self.unref_alias(alias)
# The path.join_field is a Rel, lets get the other side's field
join_field = path.join_field.field
# Build the filter prefix.
paths_in_prefix = trimmed_paths
trimmed_prefix = []
for name, path in names_with_path:
if paths_in_prefix - len(path) < 0:
break
trimmed_prefix.append(name)
paths_in_prefix -= len(path)
trimmed_prefix.append(join_field.foreign_related_fields[0].name)
trimmed_prefix = LOOKUP_SEP.join(trimmed_prefix)
# Lets still see if we can trim the first join from the inner query
# (that is, self). We can't do this for:
# - LEFT JOINs because we would miss those rows that have nothing on
# the outer side,
# - INNER JOINs from filtered relations because we would miss their
# filters.
first_join = self.alias_map[lookup_tables[trimmed_paths + 1]]
if first_join.join_type != LOUTER and not first_join.filtered_relation:
select_fields = [r[0] for r in join_field.related_fields]
select_alias = lookup_tables[trimmed_paths + 1]
self.unref_alias(lookup_tables[trimmed_paths])
extra_restriction = join_field.get_extra_restriction(
None, lookup_tables[trimmed_paths + 1]
)
if extra_restriction:
self.where.add(extra_restriction, AND)
else:
# TODO: It might be possible to trim more joins from the start of the
# inner query if it happens to have a longer join chain containing the
# values in select_fields. Lets punt this one for now.
select_fields = [r[1] for r in join_field.related_fields]
select_alias = lookup_tables[trimmed_paths]
# The found starting point is likely a join_class instead of a
# base_table_class reference. But the first entry in the query's FROM
# clause must not be a JOIN.
for table in self.alias_map:
if self.alias_refcount[table] > 0:
self.alias_map[table] = self.base_table_class(
self.alias_map[table].table_name,
table,
)
break
self.set_select([f.get_col(select_alias) for f in select_fields])
return trimmed_prefix, contains_louter
def is_nullable(self, field):
"""
Check if the given field should be treated as nullable.
Some backends treat '' as null and Django treats such fields as
nullable for those backends. In such situations field.null can be
False even if we should treat the field as nullable.
"""
# We need to use DEFAULT_DB_ALIAS here, as QuerySet does not have
# (nor should it have) knowledge of which connection is going to be
# used. The proper fix would be to defer all decisions where
# is_nullable() is needed to the compiler stage, but that is not easy
# to do currently.
return field.null or (
field.empty_strings_allowed
and connections[DEFAULT_DB_ALIAS].features.interprets_empty_strings_as_nulls
)
def get_order_dir(field, default="ASC"):
"""
Return the field name and direction for an order specification. For
example, '-foo' is returned as ('foo', 'DESC').
The 'default' param is used to indicate which way no prefix (or a '+'
prefix) should sort. The '-' prefix always sorts the opposite way.
"""
dirn = ORDER_DIR[default]
if field[0] == "-":
return field[1:], dirn[1]
return field, dirn[0]
class JoinPromoter:
"""
A class to abstract away join promotion problems for complex filter
conditions.
"""
def __init__(self, connector, num_children, negated):
self.connector = connector
self.negated = negated
if self.negated:
if connector == AND:
self.effective_connector = OR
else:
self.effective_connector = AND
else:
self.effective_connector = self.connector
self.num_children = num_children
# Maps of table alias to how many times it is seen as required for
# inner and/or outer joins.
self.votes = Counter()
def __repr__(self):
return (
f"{self.__class__.__qualname__}(connector={self.connector!r}, "
f"num_children={self.num_children!r}, negated={self.negated!r})"
)
def add_votes(self, votes):
"""
Add single vote per item to self.votes. Parameter can be any
iterable.
"""
self.votes.update(votes)
def update_join_types(self, query):
"""
Change join types so that the generated query is as efficient as
possible, but still correct. So, change as many joins as possible
to INNER, but don't make OUTER joins INNER if that could remove
results from the query.
"""
to_promote = set()
to_demote = set()
# The effective_connector is used so that NOT (a AND b) is treated
# similarly to (a OR b) for join promotion.
for table, votes in self.votes.items():
# We must use outer joins in OR case when the join isn't contained
# in all of the joins. Otherwise the INNER JOIN itself could remove
# valid results. Consider the case where a model with rel_a and
# rel_b relations is queried with rel_a__col=1 | rel_b__col=2. Now,
# if rel_a join doesn't produce any results is null (for example
# reverse foreign key or null value in direct foreign key), and
# there is a matching row in rel_b with col=2, then an INNER join
# to rel_a would remove a valid match from the query. So, we need
# to promote any existing INNER to LOUTER (it is possible this
# promotion in turn will be demoted later on).
if self.effective_connector == OR and votes < self.num_children:
to_promote.add(table)
# If connector is AND and there is a filter that can match only
# when there is a joinable row, then use INNER. For example, in
# rel_a__col=1 & rel_b__col=2, if either of the rels produce NULL
# as join output, then the col=1 or col=2 can't match (as
# NULL=anything is always false).
# For the OR case, if all children voted for a join to be inner,
# then we can use INNER for the join. For example:
# (rel_a__col__icontains=Alex | rel_a__col__icontains=Russell)
# then if rel_a doesn't produce any rows, the whole condition
# can't match. Hence we can safely use INNER join.
if self.effective_connector == AND or (
self.effective_connector == OR and votes == self.num_children
):
to_demote.add(table)
# Finally, what happens in cases where we have:
# (rel_a__col=1|rel_b__col=2) & rel_a__col__gte=0
# Now, we first generate the OR clause, and promote joins for it
# in the first if branch above. Both rel_a and rel_b are promoted
# to LOUTER joins. After that we do the AND case. The OR case
# voted no inner joins but the rel_a__col__gte=0 votes inner join
# for rel_a. We demote it back to INNER join (in AND case a single
# vote is enough). The demotion is OK, if rel_a doesn't produce
# rows, then the rel_a__col__gte=0 clause can't be true, and thus
# the whole clause must be false. So, it is safe to use INNER
# join.
# Note that in this example we could just as well have the __gte
# clause and the OR clause swapped. Or we could replace the __gte
# clause with an OR clause containing rel_a__col=1|rel_a__col=2,
# and again we could safely demote to INNER.
query.promote_joins(to_promote)
query.demote_joins(to_demote)
return to_demote
|